SCF Not Converged !! Any sugestions ?

[Amol Patil]

This time for my polyatomic system the calculations went on for almost about 589 iterations and at the end I got all the output files, however xmo file contains a flag and warning as follows.

IFLAG= -1
 LINE SEARCH FAILED. SEE DOCUMENTATION OF ROUTINE MCSRCH
 ERROR RETURN OF LINE SEARCH: INFO=  3
 POSSIBLE CAUSES: FUNCTION OR GRADIENT ARE INCORRECT
 OR INCORRECT TOLERANCES
                589       -478.0348355622     -0.0000000000      0.0066223158
 
                        VBSCF Failed to converge in   589 iterations
 
 Boys localizing
 Block                     1
 O                     1                    2                    3                    4                    5                    6                    7                    8                    9                   10                   11                   12                   13                   14                   15                   16                   17                   18                   19                   20                   21                   22                   23                   24
 X                    40
 Block                     2
 O                    25                   26                   27                   28                   29                   30                   31                   32                   33                   34                   35                   36                   37                   38                   39
 X                    41
  WARNING: The job is not convergent yet!


 ----------------------------------------------------
 WARNING: Convergence fails! Results may be wrong!
 Please try to change initial guess, or use another
 optimized method to confirm.
 ----------------------------------------------------


I am unable to figure out why? Any sugesstions?

[Amol Patil]

I take this opportunity to ask your intellectual help on some aspects.(I am rather unfortunate in not being able to register myself for the recent Valence Bond Workshop and Conference.).

I am working on protic ionic liquids. I am trying to investigate the ionicity of these ionic liquids by treating them the ab initio valence bond way so as to check the extent of proton transfer.

Whenever I make use of Hybrid Atomic Orbitals the calculations are either being killed or there is failure of optimization algorithms.I tried changing algorithms but the problem remains the same( I however did not check the efficiency of GBT algorithm and full hessian algorithm as these were too heavy for my system to cope up with.). Providing guess through is tricky as the molecule is polyatomic and the charge is highly delocalized.

Whenever I am making use of Overlap Enhanced Orbitals the calculation converges efficiently without any error flags. The molecule that I have chosen seems to show high delocalization and considerable orbital overlap.

With this effect I here I upload my Preint and XMVB 2.1 input and output files. I request you to give your insightful suggestions on this matter.

Thanking you.

[应富鸣]

Dear Amol,

Your mail has been received.

I will check it later.

Best regards,

在 2015年7月13日星期一 UTC+8下午3:37:54，Amol Patil写道：

[应富鸣]

Dear Amol,

I checked it and the probable reason was that the initial guess may not be good enough.

FRGTYP=ATOM is a relatively rough way to get localized orbitals with the default method to get initial guess. However, in this case, it may be the easiest way.

I tried the same calculation with GUESS=MO and it converged well. Attached are the xmi and xmo with GUESS=MO.

Besides, the structure

1:38 39 40 40 41

is not quite right. The orbital "40" are actually doubly occupied. So the structure should be

1:38 40 40 39 41

Best regards,

在 2015年7月13日星期一 UTC+8下午3:37:54，Amol Patil写道：

I take this opportunity to ask your intellectual help on some aspects.(I am rather unfortunate in not being able to register myself for the recent Valence Bond Workshop and Conference.).

[Amol Patil]

Thanks for the kind help !